The advent of Command Control technologies based on digital control signals has led to increased enjoyment and capabilities for model railroaders and their operations of model railroad layouts. Control technologies such as the widely used National Model Railroad Association Digital Command Control Standard (NMRA DCC) and others provide a rich set of control possibilities including the ability to control sounds in decoders in locomotive and modules on the layout. Accordingly, new generations of sound capable products that run on model railroads have been developed.
In the late 1980's Marklin GmbH of Goppingen Germany introduced an HO scale “dance car” sound-equipped coach that had sound effects controllable by function keys on their “AC Digital” digital control system. In the mid-1990's Marklin also introduced “1-Gauge” locomotives incorporating digital sounds, motor and function decoders.
Modern art sound generators suitable for use on model railroads have several broad logical elements: (a) the basic core is a decoder control means that has the electronic components and usually a data processor with associated decoder control software or firmware that can detect input voltages, external stimuli and commands and act to control any non-sound aspect such as a motor, (b) a sound control means that employs a configurable state machine or other sound sequencer control algorithm for animation of, (c) encoded sound fragments held by a sound storage means, that then may be combined by predetermined rules and output to a sound reproduction device such as a speaker. These three elements act in combination to provide an operative sound scheme that is intended to mimic some of the sounds of a real railroad.
In modern implementations these three elements are often (but not necessarily) animated by separate software algorithms that may execute on one or more cooperating data processors, and each aspect of these software implementations may benefit from the ability to be downloaded into an already installed decoder unit so as to be modified by an end user. Suitable high capability data processors with internal non-volatile storage for use in sound generators are available from many widely known integrated circuit manufacturers such as Intel Corp., Microchip Corp, Analog Devices Inc., Atmel Corp., Philips Nev., ST Corp., etc., as highly integrated function embedded control processors, microprocessors and even digital signal processors (DSP).
Novosel et al, in U.S. Pat. No. 5,855,004 teaches the benefit of digitally generated sounds in locomotive decoders when operating on an NMRA DCC control signal. This work was actually anticipated and demonstrated in the DSD2408 DCC sound decoder shown publicly by SoundTraxx, and also tested in the “1996 DCC Blowout” clinic, at the Amherst Model Train show at Springfield Mass., in February 1996, a full 14 months before being added as new matter to the Novosel '004 specification. The well-known magazine ‘Model Railroader’ in October 1996 ran on pp. 92-93 an article covering the Amherst show “1996 DCC Blowout”, that points out that in fact the product Novosel disclosed in February 1996 did not integrate motor control capability.
Novosel teaches in '004 the use of a Yamaha YM3812 integrated circuit as a digital sound generator that is followed by a YM3014 Digital to Analog Converter, or DAC, and that digitization, compression and voice-synthesis are used to convert original sound recordings into data stored in an EEPROM sound storage device that is then suitable for playback. The Yamaha device uses a well-known oscillator-based Frequency Modulation synthesis technique and is employed in most sound cards on the IBM PC's and compatibles, or sound cards like the “Sound Blaster”, “AdLib” and others.
However, Novosel fails to teach an operative method of converting general locomotive or railroad sound recordings into digital data suitable to drive the Yamaha YM3812 device.
In fact, the best approximation would be to make the Yamaha device operate as e.g. a MIDI synthesizer with multiple FM “voices”, which do not sound like any locomotive or real world sound recordings, but would render the sounds as an approximation similar to a MOOG keyboard type synthesizer. Thus the Novosel embodiment and teaching fails to provide enough information on algorithms and procedures needed to yield an operating digital sound generator, as claimed, that is suitable for playing back locomotive or model railroad sounds.
The DAST analog sound storage chip embodiment taught by Novosel in '004 is capable of storing reasonable quality arbitrary sound recordings, but suffers from the limitation noted by Novosel that it only has a single “voice” or sound channel, whereas a realistic locomotive sound synthesis requires more than one sound channel in operation at the same time. Novosel's “Real Rail Effects” product demonstrated at the Amherst show at the same time as the Soundtraxx DSD2408 in February 1996, was based on the analog DAST device.
Additionally, Novosel fails to teach or anticipate in '004 the capability of decoder downloadable sound files in for example the popular windows “.wav” format, and that these downloaded sounds can be sequenced by a modifiable control data block or file.
The Soundtraxx DSD2048 is a sound decoder that operates on NMRA DCC track control signals with digitally stored sound waveform fragments, using a parallel DAC such as an Analog Devices AD574 device to convert digital data from the sequenced sound fragments into a signal that can be amplified and then drive a speaker. The Soundtraxx device does not need the extra complexity of compressed or companded sound information, or voice synthesis to store sound data in an onboard Flash memory storage device taught by Novosel. The Soundtraxx DSD2048 also demonstrates multiple digital “voices” or simultaneous sound generator channels that are digitally mixed and then conveyed to a single speaker. Thus the prior art demonstrated and operating before Novosel, included digital “multi-voice” sound generation that is responsive to the locomotive speed and state, motor control and function control of e.g. lights and special effects, and is in every respect a superior and operative technology.
In early 1999 ESU GmbH demonstrated a DCC compatible sound decoder that allowed the user to download to the decoder in the locomotive customizable sound fragments and a control sequencing scheme from a recorded computer wave sound file (file type “.wav”, etc), CD or the Internet via the track. Information can also be uploaded from the decoder and sent via the Internet or other means for Customer Service use. The predefined operating state changes of the ESU decoder allowed the user to custom configure these looped sound fragments for steam chuffs (or diesel prime mover) when the decoder changes from accelerating under load, running steadily and decelerating with lighter load. For diesels, the prime mover pitch and volume are modulated based on throttle demand, and unique startup and shutdown sounds are also part of the sound-sequencing scheme. Note that the ESU decoder allows the decoder set up to be modified or downloaded into Flash or EEPROM memory when the locomotive is sitting on a programming track, so that it does not need to be removed to be set up. In contrast to the Soundtraxx AD574 DAC means to convert digital sound data to an analog signal to be amplified on to the speaker, the ESU art employs a Pulse Width Modulation (PWM) method with a following low-pass filter to perform this function. This is a more cost effective and compact art.
This was a big improvement over the Soundtraxx digital sound art, in that any ESU decoder stocked by a retail shop could be customized and downloaded in about 15 minutes to any of the many available locomotive sound schemes by simply hooking it up to a sound programmer. This allows a reduction in the number of decoder retail SKU's or part numbers that need to be stocked by retailers. The Soundtraxx units have dozens of sound variations predefined by the manufacturer, and so all these expensive decoder variations have to be stocked if rapid customer satisfaction is desired.
The original ESU control file that regulates sound sequencing choices that underlay the state machine that controls sound generation process were somewhat limited, and only two “voices” were generated at the same time, so in 2004 ESU introduced their version 3 decoders that allow more voices and have a more complex state machine to define the control of sound sequencing and more user configuration choices. In the ESU art this control file is a compilation of encoded binary data that is proprietary and trade secret. The ESU control file is encoded information for a state machine that some users have been able to partially decode and post some conclusions on the Internet.
However users cannot fine-tune or adjust major sound control parameters beyond the limited set of adjustments provided in the ESU programming and set up software, so in essence the heart of the ESU product capability is shielded from adaptation by a user wishing to modify the sound generator beyond the preset capabilities. This means that it is not possible to have a universal sound generator or user definable capabilities because the native processing capability of the sound generator is not fully accessible and is lacking in key capabilities.
Severson in U.S. Pat. No. 5,832,431 teaches a sound technology employing ‘pseudo-random’ or ‘random’ techniques to loop a limited set of digitally stored sound fragments so as to be perceived as playing in a non-repetitive manner. Additionally Severson '431 suggests the use of a simple “FORTRAN-like” language for the ‘random’ control sequencing of sound fragments. In the model railroad context of mimicking the sounds of parts of a full-scale railroad, the Severson '431 concept of random sounds in not very useful. In fact the physics of the operations of a railroad is in fact not random, and the sound sequences and noises are in fact linked and directly correlated to actions taken by workers on a railroad or physical characteristics and processes of the railroad equipment. For example, the rail joiner “clickety-clack” sound is wholly related to rolling-stock speeds, axle counts, locations of rail joints and the deterministic routes taken by trains etc., moving on a railroad. Horn sounds and bell ringing are the results of engineers following the railroad rule-book requirements, such as horn blasts approaching grade crossings and ringing the locomotive bell when moving within the limits of a yard.
Even the prime-mover sounds of for example a diesel engine are not ‘random’, but strongly and directly correlated to a physical process. As a diesel engine operates, one of the strongest cyclic noises is the crank-shaft rate repetitive noises such as; tappets, imbalance vibrations, engine cylinder firing cadence that then has an overlay of exhaust and/or supercharger noises. At idle, the diesel exhaust noise is at a lower level, so one may start to hear the underlying ‘hunting’ of the crank-shaft speed due to the finite diesel fuel governor control response and delay, and this is accentuated when the load on the diesel changes such as when the air compressor operates.
The air compressor is primarily used for the train brake systems and it also does not come on at ‘random’ times. Air compressor cycling is exactly defined by an air pressure-sensor detecting that air consumption from the reservoir due to braking and any leaks has reached a lower pressure limit. Thus air compressor sounds are triggered or governed by the work the locomotive and train may be performing under the control of the train engineer and conductor and the state of maintenance of the equipment. Certainly none of this is ‘random’. So in fact a ‘random’ model is inadequate to describe and control sound sequences a real railroad or a model railroad sound generator, when the sounds are truly correlated with actions commanded by operators, conductors or engineers within a scheduled work rhythm.
Some sound events that are directly correlated to current actions, such as a diesel idle being defined by an engineer choosing low throttle setting and the traction motors not engaged, may still be further modified by other influences that appear to ‘scatter’ the nature of the sounds. For this diesel idle example, the ‘scatter’ of the engine hunting or even surging while at idle is directly related to the; maintenance, immediate and long term history of the locomotive, including the lube oil lubricity and viscosity, engine wear and temperature and governor settings. Mechanical fuel governors need a finite engine speed or load change before they automatically modify the fuel flow injected into each cylinder. The energy or ‘cetane rating’ of diesel fuels also changes with age and batches, so the governor flow rates are perturbed simply when older fuel is used, etc. So a ‘scatter’ mechanism is distinctly not ‘random’, in the context of this invention, but is directly related to a knowable physical effect that can be predicted and modeled to any arbitrary precision with one or more contributing phenomena. Once a physical phenomena is characterized, then the related sounds can be created and modified so that they follow the defined scattering mechanism without requiring the concept of random or pseudo-random processes or Gaussian distributions etc., which are not accurate sound predictors and thus inferior for this application. The renowned National Institute of Standards and Technology (NIST) teaches in a published technical reference: http://www.itl.nist.gov/div898/handbook/eda/section3/scatterp.htm that a process that is considered ‘scattered’ shows “ . . . non-random structure.” in associations and causality of events.
All this points out that the art taught by Severson '431 may be useful for synthesizing continuous pink noises of an e.g. waterfall or waves on a seashore beach, but is not realistic for a working railroad sound simulation because fundamentally the vast bulk of sounds are directly related and correlated to human activities and equipment history not, ‘random’ events.
Interestingly on a model railroad, if a human can control the locomotives and any animation of the layout, then we have an exact model and trigger for sound generation and synthesis, since a human activity fundamentally lies at the basis of all the resulting sounds, just as in the prototype or real railroad equipment.
The language idea suggested by Severson '431 for defining sound sequencing algorithms has merit, but useful extensions and new concepts to provide users convenient or straightforward access to this capability are not taught or claimed. In fact model railroad sound decoders and generators from QSI Industries of Portland Oreg., that employ aspects of Severson '431 art are not sound downloadable and cannot have their sound sequencing algorithms modified by an end user, except for a limited set a adjustments allowed by predefined NMRA CV programming. These QSI decoders may have the decoder software, sound control software and/or sound files changed, since they are resident in a socketed Flash memory chip. However this is not state of the art, and means that the locomotive and static sensitive decoder must be opened up and worked on to upgrade or change the sound effects, or even fix software errors or ‘bugs’. QSI does not provide the information and examples for end users to generate their own sounds and control sequencing algorithms, so like the Soundtraxx units, users are limited to what the manufacturer explicitly provides.
The prior art does not provide for a sound generator or decoder that is fully customizable and downloadable by the end user. This lack of end user programmability of the sound sequencing and fine control of the sound sequences is not addressed by the prior art, and only inadequate or crude mechanisms have been provided for the user to modify sound generation. For example on ESU and Soundtraxx decoders it is possible to program NMRA Configuration Variables (CV's) to adjust the engine speed responses, volume levels etc, but not allow a dynamic change of throttle set points for e.g. diesel notching or governor set points to be changed by conditional user input to such items as train length and weight etc. To allow more complex and even user definable capabilities, a new art is required.
The goal of all these technologies is to create sound generators or decoders that are adaptable to the desired and arbitrary sound characteristics defined by a user. To be maximally useful and customizable, these units must be downloadable and user configurable for all the sounds and sequencing algorithms.
The provision of a capability that provides user selectable, editable and downloadable sounds and user definition and configuration of sequencing algorithms, without the aforementioned limitations of prior art, is a valuable addition to and improvement over the prior art of model railroad sound generation.